Reference · all nodes
The winkComposer node index
Every building block grouped by role, with what each one does.
Arithmetic between fields
Cumulative running total of a field.
Difference between two fields on the same message.
Invert a boolean field — the standard adapter between detection and persistence.
Divide one field by another with safe handling of zero and invalid inputs.
Apply a pure function to each sample — custom arithmetic without writing a node.
Detection and thresholds
Page-Hinkley change detector; fires when a running mean drifts beyond threshold.
Confirm a condition holds for N of M samples before firing — the standard noise filter.
Compute Cpk/Ppk process capability from mean and stdev fields against spec limits.
Flag values above, below, or inside a bound, with optional hysteresis.
Detect significant departure of a signal from a projected trajectory.
Feature extraction
Expand a compact digest into full statistics — mean, variance, skewness, kurtosis.
Track how long a condition has been active — duty cycles, state lifetimes.
Exponentially-smoothed Pearson correlation between two fields.
All pairwise correlations across a field group — dense coupling in one pass.
Compare the current value with a lagged value or the same field one message back.
Compact statistical digest over a window — pass downstream without materialising samples.
Detect transitions between symbolic states with debounce against spurious flips.
Report each completed swing — dip or peak — with its size, measured over a sliding window.
Logical reduce over a set of flag fields — any, all, or count.
Classify a signal as stable, rising, or falling from its recent slope.
Spread across fields that should read alike — NEMA percent unbalance.
Distance between two fields — RMS, cosine, angular, MAD, or maximum.
Flow control
Drops messages that fail a condition — the standard gate for downstream logic.
Intelligence and reasoning
Fuse multiple signals into one verdict — evidence builds per signal and fades as the signal returns to its normal range.
Scalar Kalman filter — model-based state estimation with outlier detection.
Observability and I/O
Broadcast a copy of the message to an external system when a condition is met.
Writes to storage when a condition is met.
Orchestration and control
Enable, disable, reset, pause, and flush nodes on conditions — the runtime’s conductor.
Signal conditioning
2nd-order low-pass or high-pass filter for frequency-domain shaping.
Map values to named categories against a list of boundaries.
Exponentially-smoothed mean; fast to respond, resistant to single-sample noise.
Smoothed mean, stdev, envelope, and zScore in one pass.
Convolutional windowed operator — smoothing, derivatives, and shaped detectors.
Three-sample median filter — cheap, effective spike removal.
Validate inputs against ranges, allow-lists, and custom checks.
Detect and clean isolated spikes while preserving legitimate step changes.
Sliding-window exact statistics over a configurable history.
Tumbling-window statistics — one emission per complete window.